The proposed research will investigate how non-pecuniary factors potentially influence the retirement decision. The key areas of focus will be caregiving responsibilities and attitudes. Existing literature documents the effects of financial incentives on the retirement decision (e.g., pensions, Social Security); one goal of the proposed research is to contribute to the (relatively new) literature that incorporates "non-economic" factors. The proposed research will extend the retirement, caregiving, and intentions literatures of endogenous decision making. Another facet of the proposed research will incorporate expectations and uncertainty into retirement models. These goals are made feasible by improved data sources and new econometric techniques. The research will proceed as follows: 1) consider the differences in attitudes and caregiving responsibilities across demographic and gender groups, 2) examine the relationship between subjective probability assessments and actual outcomes, noting that the "overall" attitude of the individual will affect these assessments, 3) investigate the endogeneity between the caregiving and retirement decisions, 4) modify a dynamic "base case" behavioral model of retirement to incorporate these non-pecuniary influences, 5) simulate the effects on the retirement decision of changes in public and corporate policy. The principal dataset to be used in the analysis is the HRA (HRS/AHEAD) survey. The HRA is a natural choice for the analysis because of its large sample of retirement-age individuals. It also contains a rich collection of financial information at the individual level, enabling models of the effects of non-pecuniary incentives to control for the financial effects which have been shown in previous studies to be quite important. The dataset also contains a set of subjective probability questions which can be used to predict outcomes in subsequent waves of the surveys.